Analisis Sentimen Berbasis Aspek pada Ulasan Aplikasi Zenius Menggunakan Algoritma Naïve Bayes

Abstract

The development of technology that provides an online educational platform is expected to support students in online learning activities. Education Platforms are needed to provide learning resources that can be accessed by students easily. Zenius is one of the learning accesses that are widely used by students. Students and users who use this application often give statements or reviews, both negative and positive on the performance of the Zenius application. From the reviews submitted by users, the company needs to know the advantages and disadvantages of the application, so that it can be an evaluation to improve the performance of the Zenius application. Therefore, this study aims to extract useful information from user sentiment on every aspect of the review, so that the information obtained will be easier and clearer. This study used a total of 1500 data and user reviews taken from the google Play Store scraping results. The plotted data will be carried out data cleaning in the preprocessing process. After that each word will be weighted using the TF-IDF feature. After the word weighted, the classification of aspect-based sentiment analysis uses the Complement Naïve Bayes algorithm which is commonly used in unbalanced data conditions. The evaluation results are presented in the confussion matrix evaluation and get an average accuracy result based on five product aspects of 80%.